A MULTI-LAYER 3D DEVICE AND METHOD OF MANUFACTURING
FIELD AND BACKGROUND OF THE INVENTION
The present invention relates to three-dimensional (3D) assembling technologies and, more particularly, to the fabrication and assembly of multi-layer 3D Micro-Electro-Mechanical Systems (MEMS) devices.
There are several conventional methods for assembling 3D semiconductor devices. Those methods were invented for packaging semiconductor Multi-Chip Modules (MCM) to generate Very-Large- Scale Integrated (VLSI) semiconductor dice, which are usually housed in semiconductor packages.
Many patents, including US patents US 5 495 398, US 5 291 061, US 6 080 264, and US 6 049 975 describe those conventional methods for MCM packaging.
3D MCMs are used to reduce the footprint (or real estate) of a circuit, increase speed, reduce noise, etc. These packages are limited to only a few layers.
These conventional methods have limitations in assembling a multi-layer MEMS device wherein the 3D structure is required for the functionality of the device, such as in 3D optical switches.
The 3D structure of an optical switch is, itself, part of the device; light may pass via the top layer through all other layers to the bottom one, or may be switched by any layer of the device. Therefore there are several requirements of such a 3D structure, including but not limited to:
• The pitch between adjacent layers has to be precisely fixed;
• The layers have to be parallel to each other; and
• Free space is needed between layers.
There is thus a widely recognized need for, and it would be highly advantageous to have a method of 3D packaging that meets the needs of multi-layer MEMS devices such as 3D optical switches.
SUMMARY OF THE INVENTION
According to the present invention there is provided a three-dimensional micro-electromechanical system (MEMS) apparatus including: a plurality of stacked substantially planar layers including active and inactive elements of the micro-electromechanical system, each layer having at least one wing for supporting said layer; at least one supporting wall for supporting the plurality of planar layers in a fixed relation to one another and to at least one supporting wall, each supporting wall having: a reference surface; and a plurality of positioning devices for supporting a respective wings, each positioning device: having at least one locating surface for positioning the wings to a specified precision relative to the reference surface; and a plurality of securing elements for holding the wings firmly in respective positioning devices.
According to further features of the invention described below there is provided a method of fabricating a three-dimensional micro-electromechanical system apparatus including the steps of: fabricating a plurality of stacked substantially planar layers comprising active and inactive elements of the micro-electromechanical system, each layer having at least one wing; fabricating at least one supporting wall for supporting the plurality of planar layers in a fixed relation to one another and to at least one supporting wall, each supporting wall having: a reference surface; and a plurality of positioning devices for supporting a respective wing, each positioning device having at least one locating surface for positioning the wings to a specified precision relative to the reference surface; fabricating a plurality of securing elements for holding the wings firmly in respective positioning devices; assembling the three-dimensional micro-electromechanical system apparatus; and securing the planar layers to the supporting walls.
According to still further features in the described exemplary embodiments there is provided a three-dimensional micro-electromechanical system (MEMS) apparatus having a plurality of
stacked substantially planar layers including active and inactive elements of the micro- electromechanical system, including: a first layer, having at least one activated-mirror optical switch and having at least one free optical path that is interruptable by the optical switch; at least one other layer having at least a first optical element connecting to a first optical fiber operative to transfer light from the first optical fiber to the at least one free optical path on the first layer; and at least one other optical element connecting to a second optical fiber residing on any layer; wherein an optical beam coming from the first optical fiber via the first optical connecting element is directed to at least one free optical path, and wherein the optical beam, upon being interrupted by the optical switch, is switched to at least one other optical element and, therethrough, to the second optical fiber.
The present invention discloses an innovative approach to design of 3D microelectro- mechanical devices, as applied to switching within optical circuits. By so doing, the present invention successfully addresses the shortcomings of presently known configurations in providing accurate alignment of optical elements by transferring accuracy provision from the assembly process to the design and manufacturing stages of components of the apparatus.
More specifically, the present invention is of an apparatus and a method suitable for a 3D MEMS device wherein features of the 3D structure contribute to functionality, and a method for assembly thereof. Features like but not limited to: free space between layers, presence of gas or liquid between layers, accurate and constant pitch between layers, accurate distance between a reference surface and corresponding surfaces of the layers, parallelism of layers.
Specifically, the present invention includes a plurality of Supporting Walls (SW) and a plurality of layers. The SWs include positioning devices to position the layers in relative to one another and to the SWs to a desired degree of precision. Active and inactive elements of the 3DD may be located on the SWs or the layers.
The positioning devices include: recesses, stairs, slots, holes, wells, stages (like rungs in a ladder), protrusions, etc, or any combination thereof for precisely locating each layer with respect to each SW.
The positioning devices contain locating surfaces, manufactured to a required degree of precision as regards location and parallelism to the reference surface, in order to position layers to a required degree of precision.
A feature of the present invention is the use of the SWs also for bearing interconnections among the various layers and between the 3DD and the external world. Interconnections may be like but not limited to electrical, fluidic, and optical.
Another feature of the present invention is use of the SWs and/or the layers for connecting optical devices on the various layers to external optical fibers by using integrated optics or hybrid optics.
A further feature of the present invention adds hybrid or integrated circuitry, in addition to conductive lines. The circuitry may control the operation of devices on the layers and act as an electronic interface between devices on the layers and the system controller, which can reside in a host computer. Devices on the layers can include: MEMS, electronic circuitry, optical devices, or any combination thereof. The advantage of this feature is that the 3DD becomes a MEMS system-on-a-package.
Another aspect of the present invention is that the SWs afford physical protection to the 3DD.
The SWs of a exemplary embodiment of the present invention may be made from any of the range of materials used in MEMS, such as a single-crystal silicon wafer, silicon on insulator (SOI) wafer, GaAs wafer, quartz, crystalline Ge, GaP, InP, SiC, metal, polymer etc.
The present invention may use any of several securing devices for securing a layer in a fixed position relative to the apparatus by forcing and holding each layer against at least two locating surfaces of a respective positioning device, including use of urging elements like but not limited to: a spring, a shape memory alloy (SMA), slopes, or bonding methods like but not limited to a ball grid array (BGA), bumps, glue, etc.
Depending on the requirements of the device the present invention may choose among several methods for manufacturing SWs. For example, when accuracy of positioning and parallelism are critical for the operation of a 3DD, an embodiment of the present invention may use a
single-crystal silicon (SCS) wafer, anisotropic etching, and photolithography masks for manufacturing SWs.
If accuracy of positioning and parallelism are less critical the present invention may use other methods including: isotropic etching, dry etching, laser ablation, etc.
If the SWs are made of metal, the manufacturing process may employ electroplating. The manufacture of polymer SWs may include using material such as Su 8 and lithographic techniques.
The number of layers in each 3DD depends on the functionality of the device and is not limited in the present invention.
Moreover, the present invention is not limited to two SWs. A 3DD with a different number of SWs may be manufactured according to the present invention.
Other objects, features, and advantages of the present invention will become apparent in the following detailed description of different embodiments and in the claims. The present invention is not limited to the embodiments described hereunder; those skilled in the art will see that many variants are possible.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention is herein described, by way of example only, with reference to the accompanying drawings, wherein:
Figure 1 illustrates a typical staired supporting wall (SW);
Figure 2 is a plan view of a typical layer of a 3D device with staired supporting walls;
Figure 3 shows a front view (Figure 3a), a side view (Figure 3b), and top view (Figure 3 c) of an exemplary embodiment of a three-dimensional device (3DD) with 5 layers, wherein the SWs have supporting stairs;
Figure 4 illustrates an exemplary embodiment of a 3DD with 5 layers, wherein the SWs have holes — front view (Figure 4a), side view (Figure 4b) and top view (Figure 4c);
Figure 5 illustrates several methods for bonding layers to the SWs;
Figure 6 shows another embodiment of a 3DD according to the present invention, using a single SW — front view (Figure 6a) and side view (Figure 6b);
Figure 7 illustrates another embodiment of the present invention having three SWs;
Figure 8 illustrates an application of the present invention in an optical switch; Figure 8a is a prior art optical switch, for comparison and Figure 8b is an optical switch with the same functionality as the prior art but manufactured according to the present invention.
The purpose of the drawings is to describe embodiments of the present invention and not for production. Dimensions of components and features shown in the figures are therefore chosen for convenience and clarity of presentation and are not necessarily to scale. For convenience, different drawings of a 3D device (3DD) may emphasize different aspects of the manufacturing process from other views of the same 3DD. For convenience, only some elements of the same type may be labeled with numerals.
DESCRIPTION OF EXEMPLARY EMBODIMENTS
The present invention is of an apparatus suitable for a 3D MEMS device wherein features of the 3D structure contribute to functionality, and a method for assembly thereof.
The principles and operation of multi-layer MEMS 3D devices and the production thereof according to the present invention may be better understood with reference to the drawings and the accompanying description.
Referring now to the drawings, in which like numerals refer to like parts throughout the several views, the embodiment of Figure 1 shows a SW 110F of a 3DD wherein the
positioning devices are stairs built thereinto. A second substantially identical SW HOB is located substantially parallel to and in line with SW 110F, as shown in Figure 3b. (Since SWs 110F and HOB are substantially identical, they will be referred to collectively as 110.)
SW 110 has a base surface 112, which serves as a reference surface for the 3DD, and two sets, left (L) and right (R), of four stairs, 114L, 116L, 118L, and 120L and 114R, 116R, 118R, and 120R. The performance of the 3DD requires that the respective heights, Hi, H2, H3, and H , above base-line 112 of each stair pair (114L/114R, etc) be accurate to a desired precision and that the upper surface of each stair be substantially parallel to base-line 112 and thus to the corresponding surfaces of other stairs. The upper surface of each stair (114 to 120) serves as the locating surface for positioning respective layers. In this embodiment, SW 110 is made of a Single Crystal Silicon (SCS) wafer with (110) orientation according to Miller indexes. This orientation is selected in order to define the surface of base-line 112 and steps 114 - 120. The end walls of each stair, 113L, 113R, 115L, 115R, etc, are shown in the drawings as perpendicular to base-line 112 and thus to steps 114 - 120, but are not necessarily so. If the functionality of the 3DD requires perpendicularity, additional processing may be needed.
The width of each level, Wi - W5, and the width of each step, 114, 116, 118, and 120 (L and R), depends on the number of layers (200a, 200b, 200c, 200 , and 200e in Figure 3a), the size of the device and the number of connecting elements, electronic 160, optic 180, or other (e.g. pipes for cooling liquid, etc), to and from each stair. Each level is successively wider than a preceding lower level: Wm > Wm-ι.
For illustrative purposes, only two electronic connections 160 and one optical connecting element 180 are shown. Those connections run between base-line 112 and external surfaces 130 and 131 respectively of SW 110. Those skilled in the art will appreciate that other connections, starting from base-line 112 or from stair, 114, 116, 118, or 120, can be incorporated in the present invention.
Conductive lines 160 may be made from metal, polysilicon, etc. Each line may be terminated 162 by solder bumps, conductive pads, or the like. The technology for manufacturing conductive lines is well known in the art.
Optical path 180 includes two optical connectors, 184 and 188, and an integrated optical wave-guide 182. Optical connector 188 is connectible to an external fiber (not shown in the drawing) and may be an integrated connector manufactured as part of the manufacturing process of SW 110, such as a V-groove or any other suitable groove, or a hybrid connector attached to SW 110. Optical connector 184, at another end of wave-guide 182, transfers light from wave-guide 182 to an appropriate layer, or vice versa. Optical connector 184 may be an optical element such as a micro-lens, a diffractive optical element, a micro-prism, etc.
In another embodiment of the present invention, optical path 180 may be installed in the layers themselves and not in SW 110. This approach improves efficiency by eliminating two optical connectors 184 and their mating connectors on the layer, as well as wave-guide 182 on SW 110. In the case that optical path 180 is located on and/or within a layer, connector 184 may transfer light from wave-guide 182 to an appropriate optical device on the same layer or on another layer via free space between the layers, thereby forming a 3D spatial device.
Another embodiment of the present invention may have optical paths in both the layers and SW 110. Having an optical path on layers and/or SW 110 simplifies the processes of assembling and aligning a 3DD. The accuracy is achieved mainly during the design and the fabrication of the product and not at the assembly stage, as described below in conjunction with Figure 8.
Yet another embodiment of the present invention includes hybrid or integrated circuitry (not shown in the drawings) in addition to conductive lines 160. The circuitry may control the operation of devices on each layer and act as an electronic interface between devices and the system controller, which can reside in an external computer. The devices on each layer may be, but are not limited to: MEMS, electronic circuitry, optical, fluidic. The advantage thereby is that the 3DD becomes a MEMS system-on-a-package. Adding hybrid or integrated circuitry to SW 110 can be done prior to assembling the 3DD by using technology commonly known to those skilled in the art.
Figure 2 is a plan view of a typical layer 200 of a 3DD with SWs 110. Layer 200, which has at least one planar surface, is preferably manufactured from a silicon wafer, although other
suitable substrate materials having a generally planar surface can be used. By way of example, quartz, glass, metal, or polymeric materials may be used to form the substrate. To achieve the best performance from the present invention it is preferable to assemble each layer so that a surface bearing sensitive devices faces the steps. In applications wherein sensitive devices are located on both surfaces of a layer, both surfaces should be substantially parallel to each other.
Layer 200 has a main area 210 (shown as unshaded) and two protruding wings, 220F and 220B (shaded). Main area 210 may contain devices (not shown in the drawing), which may be MEMS, electronic circuitry, optical, inter-connections, etc. Main area 210 lies between two substantially parallel SWs 110 while wings 220F and 220B lie upon an appropriate stair pair thereof and hold layer 200 in position.
A length, MLm, of main area 210 of a layer m (1 < m < number of layers) depends on the functionality of layer m; a width, MWm, of main area 210 layer m is limited by a distance between two SWs 110. A distance DWra between the outer edges of wings 220F and 220B is designed to support layer 200 on SWs 110. A wing length, WLm, is less than distance m and greater than the distance between the members of preceding stair pair m-1, that is Wm-1 < WLm < Wm+1. Thus, for example, length WL of the wing of fourth layer m is greater than W3 but smaller than W5 (Figure 1).
In general, the number and shapes of wings of a layer depend on the number and shapes of the SWs used in the 3DD. Some other examples are described in conjunction with Figures 3, 4, and 6.
Figure 3a is a front view, Figure 3b a side view, and Figure 3c a plan view of an embodiment of a 3DD 300 with five layers, 200a - 200e, wherein the SWs have supporting stairs. A number of layers 200 other than five can be stacked using the method of the present invention. In this embodiment, 3DD 300 includes two SWs HOF and HOB. SWs 110 may be made of a single-crystal silicon (SCS) wafer. In other embodiments the SWs 110 may be made of metal or polymer.
Figures 3a - 3c show layers 200a - 200e in place, hard against SW HOF at a front of 3DD 300 and SW HOB at a back 3DD 300.
Since bottom surface 230 of each of layers 200 is substantially planar and all of layers 200 are forced against planar surfaces that are substantially parallel to each other (stairs 114 to 120 and base-line 112), then layers 200a to 200e are substantially mutually parallel.
The lengths MLra of main areas of layers 200a to 200e may be different from one other; they can be shorter than a length of the SWs, as shown for layers 200a and 200e, or longer, as for layers 200b, 200c, and 200d.
Uppermost layer 200e is illustrated as before bonding. A bottom surface 230e of wings 220F and 220B of layer 200e is forced against stairs 120L and 120R and an edge 220FS of the wings is forced against right end wall 121R of stair 120R. Since a length of the wing is smaller than the distance between end walls 121L and 121R, a left edge 220FS' of the wing does not reach left end wall 121L. By this method, layer 220e is forced to a correct position in the structure.
For simplicity, no electrical or other connection lines on the SWs are shown in Figures 3.
Figure 3b is a side view of 3DD 300. It can be seen that bonding material 310 is placed on external faces of SWs HOF and HOB and below wings 220F and 220B. In other exemplary embodiments, bonding material may be placed on an internal side of SWs 110, or in any other suitable place.
Figure 3c is a plan view of 3DD 300 showing two SWs 110, wings 220B and 220F, and the main area 210e of fifth layer 200e, and the unobscured parts of fourth and third layers 210d and 210c.
SW 110 may be manufactured using photolithography and anisotropic wet etching. Anisotropic wet etching may be done using solutions such as KOH (potassium hydroxide) or TMAH (tetramethylammonium hydroxide), etc. This technology is known to those skilled in the art; more information about it can be found in Fundamentals of MicroFabrication (Marc Madou, CRC Press, 1997).
In wet anisotropic etching, etching speed varies according to crystal surface orientation. Therefore judicious selection of wafer orientation and mask alignment results in the fabrication of walls at a selected angle, with atomic spacing accuracy. For example, by using
(110) single-crystal silicon, one may fabricate a planar surface of each stair, 114, 116, 118, and 120 (L and R) and base-line 112 parallel to each other to within atomic accuracy. Thus, the accuracy of the photolithography mask influences the accuracy of the dimensions of SWs 110.
If accuracy and parallelism are less critical, other methods may be used, such as: isotropic etching, dry etching, laser ablation, etc.
The hybrid optical elements of connectors 184 and 188 may be precisely located in wells that may be etched anisotropically. Using this method provides for precise alignment and assembly of the optical elements and thus minimizes losses due to misalignment.
A method of manufacturing the 3DD may use a Die-Bonder or a "Pick and Place" machine, such as manufactured by companies like KARL-SUSS in Germany.
SWs HOF and HOB are precisely held in required relative positions by, for example, a mechanical jig.
Then first layer 200a is placed between SWs HOF and HOB by the Die-Bonder or "Pick and Place" machine, with wings 220F and 220B respectively lying on base-line 112 of each SW.
A bottom surface (usually the device-bearing surface) 230a of layer 200a is forced against base-line 112 and against a right end- wall 113R of base-line 112. While being held in position, layer 200a is bonded to SWs HOF and HOB. Bonding can be done by any of several methods, for instance by using bumps 310a and 320a (Figures 3a & 3b).
In other embodiments, in which the devices are located on an upper surface of layer 200a or on both surfaces thereof, additional treatment (for example polishing) may be needed to process both surfaces of layer of layer 200a to be parallel to each other.
Another method uses UV adhesive material for securing layer 200a in place instead of bumps 310a and 320a: UV glue is applied on a side of the SWs below the wings of the layer but not on the surface of the step. Therefore bonding of the layer to SWs HOF and HOB is made without affecting the parallelism of the layers. An appropriate UV light source (such as PC-3 or 3010-EC manufactured by Dymax Corporation, Torrington, CT, USA, illuminates glue
310a and 320a (Figures 3a & 3b), and cures the UV glue, thereby bonding layer 200a to SWs HOF and HOB.
Then second layer 200b is placed between SWs HOF and HOB with wings 220B and 220F lying on the respective members of stair pair 114L and 114R. A bottom surface 230b (Figure 3a) of layer 200b is forced against stairs 114R and 114L and against a right end wall 115R and the layer is similarly bonded to SWs HOF and HOB. The same process continues with layers 200c, 200d, and 200e. Using this method provides free spaces, 330a - 330d, between consecutive layers.
Figures 4a and 4b show another configuration, 400, of a 3DD with five mutually parallel layers, 450, 452, 454, 456, and 458, supported by the wings thereof, wherein the positioning devices are rectangular holes, instead of stairs. Five layers are shown by way of illustration; a number of layers other than five can be stacked using the method of the present invention.
The holes are manufactured using the same methods that have been described above, in conjunction with Figure 3, so that a bottom surface of each hole is substantially parallel to the bottom surface of every other hole.
The four wings of each layer are forced against the bottom edge of the appropriate holes, forcing all the layers to be parallel to each other. For example one of the pairs of wings (464aF, 464bF) of layer 454 is supported by bottom edges 424ad and 424bd of holes 424a and 424b respectively.
3DD 400 has a pair of substantially identical and substantially parallel SWs 410, 410F at a front and 410B at a back thereof (Figure 4b). In SW 410F, hole pairs: 420aF/420bF, 422aF/422bF, 424aF/424bF, 426aF/426bF, and 428aF/428bF support a front end of layers, 450, 452, 454, 456, and 458, respectively. Each hole pair 420aF/420bF etc, is arranged in a same plane that intersects SW 410F at a right angle. The planes correspond to layers 450 - 458 and are mutually substantially parallel.
Each hole has a lower surface, 424ad in hole 424aF and 424bd in hole 424bF, etc, that is made planar to a required precision and that serves as a locating surface for positioning the layer. Each right-hand hole, 420bF, 422bF, 424bF, 426bF, 428bF, has a right-hand end
surface such as 424br in hole 424bF that is made planar to a required precision. SW 410B has substantially identical holes.
The two horizontal surfaces of all the holes (for example in hole 424bF, an upper surface, 424bu and bottom surface 424bd) are substantially parallel to each other.
The end walls of each hole, e.g. 424br, are not necessarily perpendicular to the two horizontal surfaces of the holes (e.g. upper surface 424bu and bottom surface 424bd) but, for convenience, are shown that way. If it is required by the functionality of the 3DD that an end wall be perpendicular to horizontal surfaces, additional processes may be needed during manufacture.
Layers 450 - 458 are like layer 200, which has been described above in conjunction with Figure 2. The main difference is that each of layers 450 - 458 has at least two pairs of wings, 460aF and 460aB, and 460bF and 460bB, while layer 200 has only one pair. Each wing is smaller than a corresponding hole thereof. The wings may include connection lines to connect devices located on the layers to the external world or to another layer via connection lines 160 on SWs 410F and 410B.
Connection between layers may be done by various methods that are known in the art (not shown in the drawing) such as flex circuit from each layer or wire bonding. The present invention may also use conductive lines such as 160 that run on and/or within SWs 410 and connect conductive lines on the wings of the layers to the external world, or conductive lines such as 170 connecting wings of different layers 462aF and 464aF. As an example, bumps over pad 164 (Figures 4a and 4b) comiect conductive line 160 to a connecting conductive line (not shown in the drawing) on layer 450. Those skilled in the art will appreciate that other methods may be used for connecting the conductive lines on SWs 410 with conductive lines on layers 450 - 458. Another end of line 160 terminates at connector pin 166 (Figure 4b) that may connect to external circuitry (not shown in the drawing). Lines 160 can run on both sides of the SWs, on an external side 410E and on an internal side 4101 (not shown in the drawing).
SWs 410 may be used for optically connecting optical devices on each layer to the external world via an optical path 180 including optical connectors such as 184 and 188 and a wave-
guide 182 (Figure 4b). The present invention may use a similar optical path on at least one of the layers to connect that layer directly to external fibers (not shown in the drawing).
In some embodiments, SWs 410F and 410B may carry a plurality of optical paths 180 to connect optical devices (not shown in the drawing) on a layer to external fibers (not shown in the drawing). Optical path 180 starts with an optical connector 184 that conveys light from a suitable connector on the layer to a wave-guide (WG) 182. Another end of WG 182 can be connected to optical fiber lying in a V-groove, or other type of comiection 188 that can be connected to an external fiber (not shown in the drawing). Optical paths 180 may run on both sides of the SWs, on external side 410E (Figure 4) and on internal side 4101. More information about optical paths that may run also on layers, 450 - 458, has been given in conjunction with Figure 1.
Electrical connections and optical paths are described above in conjunction with Figure 1.
Another embodiment of the present invention includes hybrid or integrated circuitry (not shown in the drawings) on SWs 410F and 410B in addition to conductive lines. The circuitry may control the operation of devices on each layer and act as an electronic interface between devices on each layer and a system controller, which can reside in an external computer. Devices on each layer may be: MEMS, electronic circuitry, optical, fluidic, etc. The advantage thereby is that the 3D structure becomes a MEMS system-on-a-package. Hybrid or integrated circuitry can be added to SWs 410 prior to assembling the 3DD using technology known to those skilled in the art.
Figure 4c is a plan view of an exemplary embodiment of a 3DD with five layers, wherein SWs 410F and 410B, have holes and five layers, 450 - 458, two pairs of wings each. For convenience only upper three layers, 458, 456, and 454 are shown.
The two wings of each layer are forced against a right end of the appropriate holes. For example, wing pair (468bF, 468bB) of layer 458 is forced against right end 428br of holes 428b in SW 410F and 410B. Wings 468bF and 468bB do not contact left ends 428bl of holes 428b. The other pair of wings of each layer, e.g. 468aF and 468aB, do not contact any of the side edges of the holes wherein they reside, e.g. 428L and 428R, respectively.
SWs 410 may be made of materials like the materials of SW 110 and be manufactured using methods, like those described above in conjunction with Figure 1. An exemplary method of assembling 3DD 400 may use a Die-Bonder or "Pick and Place" machine, such as are manufactured by companies like KARL- SUSS.
The main difference in the manufacturing process is the mask needed for the photolithographic process.
SW 410F is held in a required position using any of several methods, such as a mechanical jig that holds SW 410F horizontally, with an outer surface facing downwards.
Left wings, i.e. 460aF and 460bF, of each of layers 450 - 458, the layers being vertical, are placed in respective holes i.e. wing 460aF in hole 420aF and wing 460bF in hole 420bF, and so on.
Then SW 410B is placed horizontally above layers 450 - 458 so that that the wings, in another end of the layers enter respective holes in SW 410B, i.e. wing 460aB in hole 420aB and wing 460bB in hole 420bB, and so on.
The mechanical jig holds SWs 410F and 410B and five layers, 450 - 458, in place while SWs 410F and 410B are forced towards each other as each layer is pushed down and right, so that wings 460bF - 468bF and 460bB - 468bB (not shown in the drawing), are pressed hard against bottom and the right edges of the holes, e.g. wing 464bF is forced against edges 424bd and 424br. Since the bottom edges of all the holes are mutually parallel, all the layers are forced to be parallel to one another.
Securing the layers to the SWs may be done by any of several methods. Some exemplary methods are described in conjunction with Figure 5 below.
Refer now to Figure 5, which illustrates several methods for securing a wing of a layer against a bottom edge of a hole located as shown by the encircled part of Figure 4a.
Figure 5a illustrates an embodiment that uses a Shape Memory Alloy (SMA) 530 to hold a layer against a bottom edge 520B of a hole 520 in a SW 505. The original thickness of SMA 530 is greater than a distance d between an upper surface of wing 510 and an upper edge
520U of hole 520, while a thickness of a deformed state of SMA 530 is smaller than distance d.
During securing of the 3DD, SMA 530 is in a deformed state and placed between the upper surface of wing 510 and the upper edge 520U of hole 520. Then the temperature is raised, causing SMA 530 to expand to an original size thereof and orcing wing 510 against bottom edge 520B of hole 520.
In another embodiment 520 represents a wedge used to hold a layer securely in place.
Figure 5b is a cross-section through a hole in SW 505 of an embodiment that uses solder pads 540 and 547 and a solder bump 545 to hold a layer wing 510 to SW 505. Those skilled in the art will appreciate that solder is not mandatory and other methods may be used according to the present invention e.g. an electro-plated connection may replace the solder.
In the securing stage of the assembly of the 3DD, the temperature is elevated, while urging layers against a bottom surface of respective holes, so that bonding solder 545 melts and bonds wing 510 against bottom edge 520B of hole 520.
Pads 540 and 547 can be on either surface, 505E or 5051, of S W 505, or on both sides thereof.
Figure 5c illustrates a hole in SW 505 of another embodiment using a curable glue, such as UV curable glue, to hold the layers to SWs 505. Those familiar with the art will appreciate that UV glue is not mandatory and another type of glue may be used according to the present invention e.g. epoxy, thermal curing, etc.
During the bonding stage of assembly of the 3DD, layer 510 is forced against bottom edge 520B of hole 520 and UV curable glue 550, such as manufactured by companies such as Al Technology Inc., NORLAND, ABLESTIK, EPO-TEK, DYMAX, Torrington, CT, USA etc, is applied within some of the corners between a bottom surface of wing 510 and external surface 505E or internal surface 5051 of SW 505, or on both surfaces.
Then at least one appropriate UV light source, such as manufactured by Dymax Engineering Adhesives Products, Torrington, CT, USA, EFOS, etc, provides UV light, which illuminates
surfaces of glue 550 so that the light penetrates and cures UV glue 550 thereby holding a bottom surface of wing 510 firmly against bottom edge 520B of hole 520.
Figures 5d and 5e respectively illustrate side and front views of SW 505 in an embodiment that uses springs 562 to hold the layers to SWs 505.
At least one comb 560 of springs 562 is fixed to external surface 505E of SW 505 overlapping hole 520 in such a way that an overlap of 562 exceeds a distance between an upper surface 510U of wing 510 and an upper edge 520U of hole 520.
Wing 510 penetrates the hole from the side of internal surface 5051 and pushes springs 562 outwards. Spring 562 reacts by forcing wing 510 against bottom edge 520B of hole 520. Upper surface 510U of wing 510 may have grooves (not shown in the drawings) perpendicular to external surfaces 5051 and 505E of side wall 505, each having a V cross- section to engage a respective tooth 562 of comb 560 to impede left-right motion L<→R in figure 5e.
Springs 562 may be made from any of various materials, such as metal, epoxy, etc. e.g. SU 8, a polymer which is preferably impervious to the etching solutions (e.g. KOH or TMAH).
An exemplary manufacturing process of SW 505 includes, among other stages, applying a layer of a polymer onto the silicon wafer. Using photolithographic techniques and an appropriate etching process, the shapes of comb 560 are generated at the appropriate locations on the wafer.
Next, holes 520 are etched by using a mask with the shapes of holes 520, and appropriate anisotropic wet etching. The etchant etches the silicon below and around comb 560 but does not etch comb 560 itself. The process results in hole 520 and associated spring 562 protruding thereover.
A further embodiment may use a hybrid comb of springs. This comb is fabricated in a separate process and later bonded to SW 505.
Yet another embodiment may use a metal comb and further utilize the metal springs to connect conductive lines on SW 505 with conductive lines on the layers.
Still another embodiment may use a shape other than a comb. For example, the spring may be a stiff, flexible plate.
Figure 5f illustrates another embodiment using a slope 572 in a transition area between the main area of 510 and wing 570 thereof. Upon pushing layer 510 against side wall 505, slope 572 acts as a wedge and generates a vertical force that secures wing 570 against bottom surface 520B of the 520.
Another configuration, 600, of the present invention, which uses a single SW 610 is shown in Figure 6. Figure 6a is a front view and Figure 6b is a side view of a single SW 3DD that, for purposes of illustration, has three layers 650.
SW 610 has three holes 620 and is manufactured as described above for SW 110 or SW 410. In order to provide the required stability to 3DD 600, SW 610 may be thicker than SWs 110 or 410. SW 610 can be made from a thick wafer, for example from a 2 mm wafer, or from several wafers bonded together to generate a wide SW.
Each layer 650 may be a different size from the others. Layers 650 are manufactured in the same way as the layers of the previous embodiments. The main difference between layers 650 and the previously described layers is that layer 650 has a single wing that may be at one of the ends thereof, like wing 662, or within the main area of the layer, like wings 664 and 666.
Figure 7 illustrates another embodiment 700 of the present invention with three SWs 710. Layers 750 of 3DD 700 have a hexagonal shape with three wings 760, which fit holes 720 in SWs 710.
Other embodiments may use other numbers of supporting walls.
Figures 8 illustrate an application of the present invention, an Optical Switch (OS). Figure 8a shows a prior art OS 800a and Figure 8b shows an OS 800b having a same functionality as OS 800a, but manufactured according to the present invention.
OS 800a is a switching device that receives a plurality of input signals from sets of input fibers 812a - 814a and provides a plurality of output signals to two sets of output fibers:
OUTPUT 1, containing fibers 812b and 814b, and OUTPUT 2, containing fibers 812c and 814c. For simplicity of presentation, OS 800a is shown having only two fibers, 812 and 814, in each port, and only two output ports OUTPUT 1 and OUTPUT 2, by way of example only. That is to say, OS 800a can have more than two fibers at each port and more than two output ports and, moreover, need not necessarily have equal numbers of fibers at each port.
OS 800a has switches, 872, 874a, 874b, and 876, arranged in a two- or three-dimensional array configuration. At least two of switches, 872 - 876, which are preferably microelectro- mechanical (MEMS) switches, reside on respective, distinct substrate layers, 852, 854, and 856.
In one embodiment, switches 872 - 876 are planar and move parallel to layers 852 - 876. In an OFF position of the switch, a light signal (represented by a beaded line) travels via a transparent zone, 862, 864a, 864b, and 866, of at least one layer. In an ON position, switches 872 to 876, which are mirrors, cover an appropriate transparent zone and reflect the light into another direction.
For example, in case that OS 800a has to transfer a light signal from fiber 814a to fiber 814b and from fiber 812a to 814c, OS 800a is set with switch 876 in an OFF position and light travels via transparent zone 866 to switch 874b which is set to an ON position and reflects light to fiber 814b. Switches 874a and 872 are set to OFF and let a second beam traverse both transparent zones 864a and 862 to reach fiber 814c.
More information about an exemplary OS 800a can be found in PCT International Publication Number WO 01/43450 A2, 14 June 2001.
Figures 8a and 8b represent a logical layer of switches in a 3D OS while the die represents a physical layer. Each logical layer has optical input/output ports (810a, 810b, and 810c). Each optical port is a one-dimensional array per logical-layer. The number of logical layers can be expanded (for example, at least one additional logical layer may be placed in a direction out of the plane of the figure). The plurality of logical layers transforms the one- dimensional array of fibers into a two-dimensional array of input and output fibers.
In general, the numbers of layers, switches, and input/output fibers of optical switches in a single device may be in the hundreds. This necessitates accurate assembly and complicated adjusting methods with spatial operating equipment to adjust the fibers in their Fiber Lens Array Plate (FLAP) and to align the FLAP within the 3DD to achieve a required optical alignment.
Figure 8b illustrates a cross-section of 3DD 800b, which is an OS manufactured according to the present invention having integrated optical connection to the fibers. OS 800b has the same functionality as OS 800a. Other optical switches with different types of mirror activation (e.g. gimbals, analog beam steering mirror, pop-up mirror, etc) may be implemented according to the present invention.
The dies thereof may be supported and positioned by a front SW and a back SW that are, however, not shown for ease of demonstrating the optical paths. The SWs may have stairs or holes as described above in conjunction with figures 3 and 4.
OS 800b connects to external fibers — a set of input fibers (812a and 814a) and two sets of output fibers (812b and 814b) and (812c and 814c) — by using direct optical coimections on the layers thereof. The optical connections include optical connectors, 188 to the fiber, an integrated wave-guide 182, and an optical connector 184 at another end of wave-guide 182. This optical path has been described above in conjunction with Figure 1. Optical connector 184 can be an optical element, such as: a micro lens, a diffractive optic element, a micro prism, etc.
To generate the same optical paths (beaded lines) as in OS 800a in this exemplary embodiment, an additional two layers, 858 and 850, have been added with optical connections to fibers 814a, 812b, and 814c and two layers 856 and 852 have been extended to include optical connections to fibers 812a, 814b, and 812c.
Optical switches 872 - 876 are set the same as in OS 800a.
The present invention solves the problem of aligning a plurality of layers in 3D space by dividing the problem into two independent processes. A first process is a 2D placement of fibers and free-space optical connectors 184 (lens, prism, etc) on a die. A second process is
the relative placement of dies. In both processes, accuracy is achieved mainly during design and fabrication of the product and not during assembly of the 3DD. Using this method eliminates the need for spatial alignment of the FLAPs. The accuracy of the position of optical path 182, 184, and 188 is determined during the design and the fabrication of the layer in the wafer and/or the die stage. The accuracy in the positioning of integrated optic elements is achieved by the accuracy of the photolithographic technique and, in case of hybrid optics, alignment is done on the wafer/die by using accurate, generally available equipment or methods, e.g. a common pick and place machine.
Another method may use V-grooves or recesses, in which the hybrid elements are placed. Those recesses or V-grooves are made by a lithographic method, with high accuracy.
The design and the fabrication process of holding the layers in their position using the supporting walls, as described above, determine positional accuracy in a direction perpendicular to the dice, including distances therebetween.
Other embodiments, which use integrated optical connection to the fibers, according to the present invention may use methods other than supporting walls to hold the layers (dice) in position e.g. strut poles, adhesive material, anodic bonding, etc.
This method of connecting fibers to OS 800b reduces the complicated alignment process and improves the accuracy in comparison to the FLAPs 810a, 810b, and 810c, of OS 800a.
While the invention has been described with respect to a limited number of embodiments, it will be appreciated that many variations, modifications and other applications of the invention may be made including some or all of the features of the disclosed invention as required for specific applications.
In the description and claims of the present application, each of the verbs, "comprise" "include", and "have", and conjugates thereof, are used to indicate that the object or objects of the verb are not necessarily a complete listing of members, components, elements, or parts of the subject or subjects of the verb.